IT SYSTEM FOR COUNTRY PROTECTION AGAINST EXTREME HAZARDS (ISOK) METEOROLOGICAL HAZARD MAPS METHODOLOGICAL APPROACH Agnieszka WYPYCH, Zbigniew USTRNUL, Ewelina HENEK & Co Budapest, 15 May 2014
THE MAIN COMPONENTS OF THE ISOK SYSTEM: LIDAR Data (4-12 points/m 2 ;>200 000 km 2 territoryof Poland) Digital TerrainModel (0,5-1 m; >200 000 km 2 territoryof Poland) Orthophotomaps (10 cm; for 203 towns in Poland) ISOK IT SYSTEM Flood hazard maps Flood risk maps THE METEOROLOGICAL HAZARD MAPS Technological hazard maps The ISOK project focuses on delivering the flood hazard and flood risk maps requiredby the EU Floods Directive (by the end of 2013). Map of Hydrographic Division of Poland (1: 10 000) The main goal of the project is to give citizens assurance that they are safe and to limit the losses caused by technological and natural disaster.
The MAIN GOAL of METEOROLOGICAL HAZARD MAPS is to estimate the hazard related to the weather extremes Temperature extremes Heavy rainfalls (affecting floods) Strong winds Intensive snowfalls Thunderstorms and hail Fogs Glaze Rime climatological analysis estimation of spatial differentiation of weather extremes current analysis hazard forecasting due to weather extremes
CHALLENGES / PROBLEMS: DATA: sparse station network spatially discontinuous phenomena (dependent on local environmental conditions) observations not measurements DATA HOMOGENEITY: data sources observations not measurements extremes DATA SPATIALIZATION: information vs. appearance reality vs. modeled world
DATA: daily resolution 1951-2010 (1966-2010) app. 60 350 stations/posts METHODS:
THE METEOROLOGICAL HAZARDS MAP HISTORICAL (CLIMATOLOGICAL) MAPS Air temperature Precipitation Snow cover Wind Maps present the selected quantiles of weather parameters as well as the events freqency Thunderstorms with hail Fog Glaze Rime Maps present the frequency of days with selected weather parameters as well as the probability of conditions favourable for their occurence OPERATIONAL (FORECASTING) MAPS Maps will present the weather-endangered regions: extreme phenomena or favourable conditions occurence (with possible hazard: very high, high, moderate and weak)
THE METEOROLOGICAL HAZARDS MAP CLIMATOLOGICAL MAPS ( 721 ) CONTOUR MAPS decades/ months SYMBOL MAPS months/ decades CONTOUR MAPS of FAVOURABLE CONDITIONS year/ season spatialisation of modeled RegCM values (definedalgorithms) AIR TEMPERATURE PRECIPITATION SNOW COVER WIND WIND SNOW COVER THUNDERSTORMS WITH HAIL FOG GLAZE RIME THUNDERSTORMS WITH HAIL FOG GLAZE RIME
SYMBOL MAPS THUNDERSTORMS with HAIL GLAZE RIME FOG Average annual number of days with fog (1966-2010)
THE METEOROLOGICAL HAZARD MAPS
before after
CONTOUR MAPS AIRTEMPERATURE PRECIPITATION SNOWCOVER WIND
CONTOUR MAPS AIR TEMPERATURE/ SNOW COVER Residual kriging (probability maps) Simple kriging (frequency maps)
- TMAX with the occurence probability of 1, 5 and 10% (for 36 decades): TMAX (p90, p95 and p99) January (1st decade) TMAX (p90, p95 and p99) July (1st decade) - TMIN with the occurrence probability of 1, 5 and 10% (for 36 decades): TMIN (p1, p5 and p10) January (1st decade) TMIN (p1, p5 and p10) April (1st decade)
CONTOUR MAPS PRECIPITATION Natural neighbour IDW (Inverse Distance Weighted) RBF (Radial Basis Functions) OK (Ordinary kriging) combined methods EBK (Empirical Bayesian Kriging)
RBF BIAS RR MAX P90 0,3 mm P95 0,5 mm P99 2,09 mm OK BIAS RR MAX P90 6,43 mm P95 7,13 mm P99 14,35 mm
CONTOUR MAPS OK - defaults OK - individuals defaults RMSE 6,9 individuals RMSE 5,9
Daily RR with the occurence probability of 1, 5 and 10% (for 36 decades): MAY 1st decade JULY 2nd decade EBK (Empirical Bayesian Kriging) geostatistical interpolation automated data model (individual parameters)
CONTOUR MAPS in-situ data* WIND modeled data* bilinear interpolation ordinary kriging *data inhomogeneity *WRF regression kriging
Modeled wind gust speed with the occurence probability of 2, 5 and 10 years
CONTOUR MAPS of FAVOURABLE CONDITIONS THUNDERSTORMS with HAIL GLAZE RIME FOG Average annual number of days with fog (1966-2010) Probability of thunderstorms-favourable conditions (1966-2010)
NCEP/NCAR reanalysis (2.5 x 2.5 ) RegCM model DOWNSCALLING ~20 km spatial resolution gridded 1966-2010 data with 3h temporal resolution
THUNDERSTORM: MUCAPE > 200 J/kg convective precipitation (for 23 isobaric levels) GLAZE: precipitation temperature on 700, 850 and 925 hpa isobaric level and near the ground (5 cm) RIME: probability of rime as a function of relative humidity and 2m air temperature lack of precipitation
FOG: visibility as a function of relative humidity lack of precipitation EXPLANATORY VARIABLES relief: elevation (m a.s.l.) topography (landform) landuse: forest / non forest water bodies urbanisation
Probability of occurence fog-favourable conditions visibility vegetation elevation water bodies TPI -Topography Position Index urbanisation
FORECASTING MAPS ALADIN mesoscale model, horizontal spatial resolution 13 km ( 7.5 km) AROME convective model, horizontal spatial resolution 2,7km
WIND SPEED (GUSTS) SNOW COVER THUNDERSTORM WITH HAIL FOGS RIME GLAZE simple exact interpolation (?)
AIR TEMPERATURE PRECIPITATION thresholds dependant on historical information 0.01
1) point to point (nearest neighbour) 2) point to points 3) points to points?
CONCLUSIONS / CONCERNS: Weather extremes rule their own rules difficult to homogenize and spatialize Defining conditions favourable for phenomena occurence seems to be the solution if spatial information is needed There is no universal interpolation / spatialization method (!) trying to automate maps production the best at the time should be chosen with the respect to map dedication
THANK YOU FOR YOUR ATTENTION!